Mark E. Mallett wrote:
On Wed, Jun 29, 2005 at 02:01:31PM -0700, Philip Guenther wrote:
B) require implementations to reject parts other than "from" and "to"
unless declared via some new extension
I'd go with this one. I've silently ignored others up until this
point but have had a long-standing private implementation note that
it needs to be resolved, vis:
TODO /* Not really clear if this is an error- but for now we'll
just treat it as a non-match or a zero count.
*/
CONS: makes Cyrus implementation non-conforming; breaks scripts that
haven't been updated to require the new extension
There are other dimensions to the question, of course. One is whether a
prohibition under (B) will be a "SHOULD NOT" or a "MUST NOT" (MUST NOT
seems better to me, but SHOULD NOT will work better as a retroactive
recommendation);
As a user of Cyrus Sieve I would obviously prefer SHOULD NOT here.
another is whether the spec should have an opinion on
whether the error should be detected at parse/compile time or at
execution time (I'd either go with either the latter or unspecified).
IMHO, my general answer to this is "as soon as possible". But if Sieve
"variables" are also implemented, it might not be possible to check this
until runtime.